Hi Bob, Brian,
I'm a bit concerned about this suggestion, in section 4.1, Routing :
(as a side note, this is from Rev 8, the nokia web site resolves to an
IPv6 address, I don't seem to be able to get to it via my 6to4
connection though)
For link-state IGPs, it is suggested that a site utilizing ULA
prefixes be contained either within one IGP domain or area. By
containing a ULA prefix to a single link-state area or domain, the
distribution of prefixes can be controlled.
I think it potentially could cause conflicts with the idea of using the
same subnet numbers as those used with a global prefix.
Most entities (organisations, individuals etc) who might use Local Ipv6
addresses may also be assigned a single global /48, as per the RFC3177
recommendation.
If an organisation follows the above suggestion, they may end up with a
number of ULA /48s, in particular if they generate a new ULA per IGP
area. Naively, they may, within the each area or IGP domain, assign
subnet numbers starting at 0x0001. In other words, within an area or IGP
domain, use a local 2^16 subnet address space.
This is probably reasonable and even preferred, as the length of time
the organisation "owns" its ULA address spaces is (likely to be) much
longer than the time they hold a global prefix. Focusing on the
available ULA address space while planning the subnet structure makes
sense.
The problem would occur when they try to assign these same subnet
numbers to a single, global /48 they've been assigned. Of course, they
couldn't, as they'd now be trying to assign multiple "subnet 0x0001"s
within the single, global /48. They'd be forced to use different subnet
numbers for their global and ULA prefixes.
If they were less naive, they might assign ranges of non-overlapping
subnet numbers to the different areas or IGP domains within the ULA
/48s, then allowing them to have unique subnets within the single,
global /48 they've been assigned. In other words, the subnet space they
assign subnets out of is a single range of 2^16 across their network.
This introduces complexity though, as they'll now have to configure
prefix summarisation on the area or IGP domain at different points
within the ULA and global address spaces. E.g., for the ULA assignments,
they can summarise at the /48 point, where as for the single global /48
assignment, they might summarise at, for example, the /56 point. It gets
a bit more complicated if they need to use variable length subnets
within their global IPv6 prefix. I don't think this is overly complex,
however I think it defeats the simplicity goals of either a ULA prefix
per IGP domain or area, or using common subnet numbers for ULA and
global prefixes.
I'm not sure what my opinion is regarding which addressing model would
be best, and maybe there isn't one "best" one. I suppose both of them
could be suggested, with the caveats of each of them described.
Possibly, it may be outside of the scope of this RFC. More discussion
might help.
Thanks,
Mark.
--
"This signature intentionally left blank."
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
